Events Calendar

Solution Methods to Problems that Involve Discrete Choice Constraints

DANIEL J. EPSTEIN DEPARTMENT OF INDUSTRIAL & SYSTEMS ENGINEERING SEMINAR:Title: "Solution Methods to Problems that Involve Discrete Choice Constraints"Speaker: Dr. Stephen Stoyan, Visiting Assistant Professor, Daniel J Epstein Department of Industrial and Systems Engineering, University of Southern CaliforniaABSTRACT: There are a number of practical problems in health care, supply-chain management, energy, and finance that involve the decision or selection of a subset of elements within the system. In each case, the problem equates to solving a mixed-integer program with discrete choice constraints. Due to an inherent NP-hard subproblem, such models are difficult to solve and commercial solvers will not provide optimal solutions. We present an iterative penalty algorithm that takes advantage of problem structure and generates solutions for practical instances. The algorithms performance is illustrated in a financial example involving portfolio selection, where the results are compared to a commercial solver and other approaches. Even greater performance is achieved for large-scale instances when uncertainty is added to the problem. Such designs can be applied to patient scheduling and supply-chain network models, where the approach represents a markedly different but effective method with respect to the literature. We show that the algorithm performs remarkably well in comparison to a well-known benchmark and solutions are generated within reasonable time frames.TUESDAY, MARCH 30, 2010, ANDRUS GERONTOLOGY BUILDING, (GER) ROOM 309, 3:30 – 4:30 PM

Location: Ethel Percy Andrus Gerontology Center (GER) - 309

Audiences: Everyone Is Invited

Contact: Georgia Lum